From patchwork Thu May 31 11:35:49 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Chris Wilson X-Patchwork-Id: 10440939 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork.web.codeaurora.org (Postfix) with ESMTP id 93849602BD for ; Thu, 31 May 2018 11:37:44 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 809C92921A for ; Thu, 31 May 2018 11:37:44 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id 75461292E7; Thu, 31 May 2018 11:37:44 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-5.2 required=2.0 tests=BAYES_00, MAILING_LIST_MULTI, RCVD_IN_DNSWL_MED autolearn=ham version=3.3.1 Received: from gabe.freedesktop.org (gabe.freedesktop.org [131.252.210.177]) (using TLSv1.2 with cipher DHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.wl.linuxfoundation.org (Postfix) with ESMTPS id DCDC12921A for ; Thu, 31 May 2018 11:37:43 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 5FB4989FD4; Thu, 31 May 2018 11:37:43 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from fireflyinternet.com (mail.fireflyinternet.com [109.228.58.192]) by gabe.freedesktop.org (Postfix) with ESMTPS id D72D289FD4 for ; Thu, 31 May 2018 11:37:41 +0000 (UTC) X-Default-Received-SPF: pass (skip=forwardok (res=PASS)) x-ip-name=78.156.65.138; Received: from haswell.alporthouse.com (unverified [78.156.65.138]) by fireflyinternet.com (Firefly Internet (M1)) with ESMTP id 11899107-1500050 for multiple; Thu, 31 May 2018 12:35:56 +0100 Received: by haswell.alporthouse.com (sSMTP sendmail emulation); Thu, 31 May 2018 12:35:55 +0100 From: Chris Wilson To: intel-gfx@lists.freedesktop.org Date: Thu, 31 May 2018 12:35:49 +0100 Message-Id: <20180531113552.13152-2-chris@chris-wilson.co.uk> X-Mailer: git-send-email 2.17.1 In-Reply-To: <20180531113552.13152-1-chris@chris-wilson.co.uk> References: <20180531113552.13152-1-chris@chris-wilson.co.uk> X-Originating-IP: 78.156.65.138 X-Country: code=GB country="United Kingdom" ip=78.156.65.138 Subject: [Intel-gfx] [PATCH 2/5] drm/i915/gtt: Read-only pages for insert_entries on bdw+ X-BeenThere: intel-gfx@lists.freedesktop.org X-Mailman-Version: 2.1.23 Precedence: list List-Id: Intel graphics driver community testing & development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" X-Virus-Scanned: ClamAV using ClamSMTP From: Jon Bloomfield Hook up the flags to allow read-only ppGTT mappings for gen8+ Signed-off-by: Jon Bloomfield Signed-off-by: Chris Wilson Cc: Joonas Lahtinen Cc: Matthew Auld Reviewed-by: Joonas Lahtinen --- drivers/gpu/drm/i915/i915_gem_gtt.c | 45 ++++++++++++++++--------- drivers/gpu/drm/i915/i915_gem_gtt.h | 7 +++- drivers/gpu/drm/i915/intel_ringbuffer.c | 11 ++++-- 3 files changed, 44 insertions(+), 19 deletions(-) diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c index 5936f0bfdd19..1ac626cacc8d 100644 --- a/drivers/gpu/drm/i915/i915_gem_gtt.c +++ b/drivers/gpu/drm/i915/i915_gem_gtt.c @@ -204,7 +204,7 @@ static int ppgtt_bind_vma(struct i915_vma *vma, return ret; } - /* Currently applicable only to VLV */ + /* Applicable to VLV, and gen8+ */ pte_flags = 0; if (vma->obj->gt_ro) pte_flags |= PTE_READ_ONLY; @@ -1008,10 +1008,11 @@ gen8_ppgtt_insert_pte_entries(struct i915_hw_ppgtt *ppgtt, struct i915_page_directory_pointer *pdp, struct sgt_dma *iter, struct gen8_insert_pte *idx, - enum i915_cache_level cache_level) + enum i915_cache_level cache_level, + u32 flags) { struct i915_page_directory *pd; - const gen8_pte_t pte_encode = gen8_pte_encode(0, cache_level, 0); + const gen8_pte_t pte_encode = gen8_pte_encode(0, cache_level, flags); gen8_pte_t *vaddr; bool ret; @@ -1062,14 +1063,14 @@ gen8_ppgtt_insert_pte_entries(struct i915_hw_ppgtt *ppgtt, static void gen8_ppgtt_insert_3lvl(struct i915_address_space *vm, struct i915_vma *vma, enum i915_cache_level cache_level, - u32 unused) + u32 flags) { struct i915_hw_ppgtt *ppgtt = i915_vm_to_ppgtt(vm); struct sgt_dma iter = sgt_dma(vma); struct gen8_insert_pte idx = gen8_insert_pte(vma->node.start); gen8_ppgtt_insert_pte_entries(ppgtt, &ppgtt->pdp, &iter, &idx, - cache_level); + cache_level, flags); vma->page_sizes.gtt = I915_GTT_PAGE_SIZE; } @@ -1077,9 +1078,10 @@ static void gen8_ppgtt_insert_3lvl(struct i915_address_space *vm, static void gen8_ppgtt_insert_huge_entries(struct i915_vma *vma, struct i915_page_directory_pointer **pdps, struct sgt_dma *iter, - enum i915_cache_level cache_level) + enum i915_cache_level cache_level, + u32 flags) { - const gen8_pte_t pte_encode = gen8_pte_encode(0, cache_level, 0); + const gen8_pte_t pte_encode = gen8_pte_encode(0, cache_level, flags); u64 start = vma->node.start; dma_addr_t rem = iter->sg->length; @@ -1195,19 +1197,21 @@ static void gen8_ppgtt_insert_huge_entries(struct i915_vma *vma, static void gen8_ppgtt_insert_4lvl(struct i915_address_space *vm, struct i915_vma *vma, enum i915_cache_level cache_level, - u32 unused) + u32 flags) { struct i915_hw_ppgtt *ppgtt = i915_vm_to_ppgtt(vm); struct sgt_dma iter = sgt_dma(vma); struct i915_page_directory_pointer **pdps = ppgtt->pml4.pdps; if (vma->page_sizes.sg > I915_GTT_PAGE_SIZE) { - gen8_ppgtt_insert_huge_entries(vma, pdps, &iter, cache_level); + gen8_ppgtt_insert_huge_entries(vma, pdps, &iter, cache_level, + flags); } else { struct gen8_insert_pte idx = gen8_insert_pte(vma->node.start); while (gen8_ppgtt_insert_pte_entries(ppgtt, pdps[idx.pml4e++], - &iter, &idx, cache_level)) + &iter, &idx, cache_level, + flags)) GEM_BUG_ON(idx.pml4e >= GEN8_PML4ES_PER_PML4); vma->page_sizes.gtt = I915_GTT_PAGE_SIZE; @@ -1614,6 +1618,9 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt) 1ULL << 48 : 1ULL << 32; + /* From bdw, there is support for read-only pages in the PPGTT */ + ppgtt->base.has_read_only = true; + /* There are only few exceptions for gen >=6. chv and bxt. * And we are not sure about the latter so play safe for now. */ @@ -2430,7 +2437,7 @@ static void gen8_ggtt_insert_page(struct i915_address_space *vm, static void gen8_ggtt_insert_entries(struct i915_address_space *vm, struct i915_vma *vma, enum i915_cache_level level, - u32 unused) + u32 flags) { struct i915_ggtt *ggtt = i915_vm_to_ggtt(vm); struct sgt_iter sgt_iter; @@ -2438,6 +2445,9 @@ static void gen8_ggtt_insert_entries(struct i915_address_space *vm, const gen8_pte_t pte_encode = gen8_pte_encode(0, level, 0); dma_addr_t addr; + /* The GTT does not support read-only mappings */ + GEM_BUG_ON(flags & PTE_READ_ONLY); + gtt_entries = (gen8_pte_t __iomem *)ggtt->gsm; gtt_entries += vma->node.start >> PAGE_SHIFT; for_each_sgt_dma(addr, sgt_iter, vma->pages) @@ -2564,13 +2574,14 @@ struct insert_entries { struct i915_address_space *vm; struct i915_vma *vma; enum i915_cache_level level; + u32 flags; }; static int bxt_vtd_ggtt_insert_entries__cb(void *_arg) { struct insert_entries *arg = _arg; - gen8_ggtt_insert_entries(arg->vm, arg->vma, arg->level, 0); + gen8_ggtt_insert_entries(arg->vm, arg->vma, arg->level, arg->flags); bxt_vtd_ggtt_wa(arg->vm); return 0; @@ -2579,9 +2590,9 @@ static int bxt_vtd_ggtt_insert_entries__cb(void *_arg) static void bxt_vtd_ggtt_insert_entries__BKL(struct i915_address_space *vm, struct i915_vma *vma, enum i915_cache_level level, - u32 unused) + u32 flags) { - struct insert_entries arg = { vm, vma, level }; + struct insert_entries arg = { vm, vma, level, flags }; stop_machine(bxt_vtd_ggtt_insert_entries__cb, &arg, NULL); } @@ -2672,7 +2683,7 @@ static int ggtt_bind_vma(struct i915_vma *vma, struct drm_i915_gem_object *obj = vma->obj; u32 pte_flags; - /* Currently applicable only to VLV */ + /* Applicable to VLV (gen8+ do not support RO in the GGTT) */ pte_flags = 0; if (obj->gt_ro) pte_flags |= PTE_READ_ONLY; @@ -3555,6 +3566,10 @@ int i915_ggtt_init_hw(struct drm_i915_private *dev_priv) */ mutex_lock(&dev_priv->drm.struct_mutex); i915_address_space_init(&ggtt->base, dev_priv, "[global]"); + + /* Only VLV supports read-only GGTT mappings */ + ggtt->base.has_read_only = IS_VALLEYVIEW(dev_priv); + if (!HAS_LLC(dev_priv) && !USES_PPGTT(dev_priv)) ggtt->base.mm.color_adjust = i915_gtt_color_adjust; mutex_unlock(&dev_priv->drm.struct_mutex); diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.h b/drivers/gpu/drm/i915/i915_gem_gtt.h index aec4f73574f4..b0929b547846 100644 --- a/drivers/gpu/drm/i915/i915_gem_gtt.h +++ b/drivers/gpu/drm/i915/i915_gem_gtt.h @@ -309,7 +309,12 @@ struct i915_address_space { struct list_head unbound_list; struct pagevec free_pages; - bool pt_kmap_wc; + + /* Some systems require uncached updates of the page directories */ + bool pt_kmap_wc:1; + + /* Some systems support read-only mappings for GGTT and/or PPGTT */ + bool has_read_only:1; /* FIXME: Need a more generic return type */ gen6_pte_t (*pte_encode)(dma_addr_t addr, diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c index 97b38bbb7ce2..6cac4ce59438 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -1111,6 +1111,7 @@ void intel_ring_unpin(struct intel_ring *ring) static struct i915_vma * intel_ring_create_vma(struct drm_i915_private *dev_priv, int size) { + struct i915_address_space *vm = &dev_priv->ggtt.base; struct drm_i915_gem_object *obj; struct i915_vma *vma; @@ -1120,10 +1121,14 @@ intel_ring_create_vma(struct drm_i915_private *dev_priv, int size) if (IS_ERR(obj)) return ERR_CAST(obj); - /* mark ring buffers as read-only from GPU side by default */ - obj->gt_ro = 1; + /* + * Mark ring buffers as read-only from GPU side (so no stray overwrites) + * if supported by the platform's GGTT. + */ + if (vm->has_read_only) + obj->gt_ro = 1; - vma = i915_vma_instance(obj, &dev_priv->ggtt.base, NULL); + vma = i915_vma_instance(obj, vm, NULL); if (IS_ERR(vma)) goto err;